By Evan Hernandez, Kelly Hepper, Julia Porter, Gabi Schwartzman, Xinxu Shen, and Yacong Wu

Introduction

It is well established that in everyday speech females tend to use more paralinguistic cues than men. That is, women more regularly use facial expressions, body language, pitch inflection, and tone (among other cues) to communicate. Nancy Briton and Judith Hall’s 1995 study supports this, finding that women are better than men at sending, receiving, and interpreting nonverbal cues during conversation. Their study also finds that the nonverbal behaviors used by women tend to be more expressive.

Unfortunately, these nonverbal cues are not available when communicating in real-time through an electronically mediated medium like texting. Texters and online communicators must thus compensate for the lack of paralinguistic cues, and there are a handful of ways for them to do so. Some examples include use of punctuation, capitalization, different spellings, and emoticons. While none of these are direct substitutions for physical paralinguistic cues, they add similar information to the electronic message.

Because there are differences in how males and females use nonverbal cues in face-to-face speech, one naturally asks if there are differences in how males and females use “electronic paralinguistic cues.” We chose to address this question in the present study by looking at gender differences in the use of emoji. We predicted, based on the findings of Briton and Hall (1995), that females are more likely to use more expressive emojis in real-time written communication.

We were first interested in whether certain emojis were perceived as more masculine or feminine when used in electronically mediated forms of communication.  We then investigated whether people make assumptions about the gender of a message sender given a specific emoji in an otherwise neutral text conversation. We predicted that messages with masculine (i.e., less expressive) emojis would more likely be interpreted as messages sent by males, and messages with feminine (i.e., more expressive) emojis would more likely be interpreted as messages sent by females. Moreover, we predicted that texters using masculine emojis would be linked with more masculine personality traits, and likewise for texters using feminine emojis.

Pilot Survey

We first wanted to determine if certain emojis are perceived as more masculine or more feminine than others. We distributed a survey to 187 participants that asked them to organize ten emojis from most feminine to most masculine.  We chose to use the ten most common emojis on twitter, taken according to the data scraped by http://emojitracker.com/. The survey was distributed through Facebook, mostly to college students.

A total of 187 respondents participated in our survey, of which 51 were males, 132 were females, and 5 did not specify their gender. We found that 😘 was rated most feminine, followed by 😊, then 😂. The most masculine emoji was 👍, followed by 👌, then 😑.  The remaining four emojis were not consistently rated as being more feminine or masculine, so they were considered “neutral” emojis in our study.

initialSurveyResults
Ranking counts by emoji. Note “answerX” means that the emoji was given ranking X.

Experiment

We found from the pilot survey that people do in fact associate certain emojis with a certain gender, and indeed masculine emojis tend to be less expressive than feminine emojis. Since the respondents of the pilot survey were in fairly strong agreement about what the top three masculine and feminine emojis were, we used the top three for each gender in the experimental portion of our study. Specifically, we embedded these emojis into otherwise neutral texts and gauged how they affected the perceived gender of the texter and the perceived tone of the text.

Our hypothesis was this: if a neutral text message contains an emoji with a gendered connotation, individuals will expect the author of the text message to be of the associated gender. Moreover, individuals will judge the tone of the text in predictable ways based on personality traits typically associated with the aforementioned gender. For example, suppose we provided the text, “are you still at work? 😘” We predicted that individuals were more likely to say this text was sent by a girl, and we also predicted that they would label its tone with a more typically feminine property (e.g., “flirtatious”).

Methods

We selected 6 emojis: three were rated as masculine in the first survey, and the other three were rated as feminine. We decided to use texting as our electronically mediated communication system. We came up with three short text conversations that were gender-neutral: one was about going out for lunch, another was about what happened at a party, and the last was about making weekend plans. Each emoji was embedded into one of the messages. There were two versions of each conversation: one version contained a feminine emoji, and the other contained a masculine emoji, which gave us three pairs of text conversations. The emojis appeared in the same position in each message to control for placement. Also, only one emoji appeared in each conversation, so as to elicit the effect of that single emoji.

We administered this experiment with a Qualtrics survey. The respondents were first asked to indicate their own gender. After that, each respondent was shown three text conversations, one from each pair. Whether the conversation was the masculine or feminine version was randomized. For each conversation, respondents were asked to rate how likely they think the texter that used the emoji was male or female.  They were then shown a list of tone descriptors and were asked to indicate which tones were conveyed in the corresponding message. These descriptors included excited, angry, flirtatious, awkward, sarcastic, annoyed, embarrassed, friendly, and other.

The survey was distributed through Facebook. Participants were not compensated. The complete list of questions can be found in the appendix at the bottom of this page.

Results

There were 233 respondents to this survey, of which 148 were female and 85 were male. Show below are the complete results.

data
Survey Results. The first two emojis were used in the same conversation, as were the following two, etc. “Gender” is on a scale from 1-4, where 1 is most masculine and 4 is most feminine.

As hypothesized, people do have perceived gender bias of certain emojis. 80% of respondents thought the message with the 😊 emoji was sent by a female (M=2.84)** and 63% of respondents thought the same message with the 👌 emoji was sent by a male (M=2.36). Similarly, 69% of respondents perceived message with 😘 to have been sent by a female (M=2.81), and 71% of respondents perceived the same message with 👍 to have been sent by a male (M=2.19). 68% of respondents perceived the message sender as female (M=2.76) when using the 😂 emoji, and a surprising 76% of respondents still thought the text was sent by a female (M=2.84) when this emoji was switched with the more masculine 😑. Broken down by gender, one finds that male respondents interpreted the messages as more masculine than female respondents, and female respondents interpreted the messages as more feminine.

The perceived tones of the first conversation were friendly and excited, which were the same in female (😊) and male (👌) versions. In the second conversation, the top three perceived tones in female version (😂) were embarrassed (54%), friendly (32%) and awkward (17%); the top three perceived tones in male version (😑) were embarrassed (59%), annoyed (41%) and awkward (19%). In the third conversation, the top two perceived tones were flirtatious and friendly in both female (😘) and male version (👍). Again, when the results are broken down by the participant’s gender, one finds no reliable differences in frequency of each tone descriptor.

**The means used reflect the rating given by each respondent. The rating scale is as follows: 1=Definitely Male; 2=Probably Male; 3=Probably Female; 4=Definitely Female. Higher means indicate that more respondents thought the message sender was female, and lower ratings indicate that more respondents thought the message sender was male.

Discussion

It is fairly clear from both our pilot survey and our experiment that there are different gender associations with different emojis, and these emojis affect the perceived gender of the texter. This directly supports the first component of our hypothesis. It is also clear from our results that gender perception becomes more polarized when the data are split by participant gender; that is, males interpret texts as more masculine in general, and females interpret texts as more feminine in general. In some cases, males even interpreted a text with a feminine emoji as being sent by a male, while females behaved as predicted, although the differences were subtle.

Our experiment also shows that the perceived gender of an emoji influences the perceived tone of the message itself. For example, about 15% of participants described the text with 👌 as friendly, while 35% described it as excited when 😊 was used instead. The more feminine emoji elicited a more feminine perceived tone. This was also the case in the message where 👍 and 😘 were swapped; when the latter was used, the vast majority interpreted the tone as flirty, while when the former was used the majority interpreted the tone as friendly. If one views flirtatiousness as a stereotypically feminine trait and friendliness as a neuter, stereotypically masculine trait, then these results align with our hypothesis.

However, this effect was not always reliable. The same tone descriptors were often selected for both conversations in the pair, and each by a comparable number of participants. One explanation for this result could be that, while certain emojis are considered more feminine or masculine than others, the tonal interpretation of an emoji is tied more closely to situation than to gender. Perhaps 😑 is considered a masculine emoji in general, but there are situations where males and females are equally likely to use it, or even where females are more likely to use it than males. We observed such a case in one of our text conversations, where we swapped the feminine 😂 with the masculine 😑. A greater percentage of participants thought that the texter was a female wheemotionlessn the text ended with 😑 than when it ended with 😂, which completely contradicted our prediction. However, when one inspects the message, one notices that it does not provide a neutral context. Instead, the text is about an incident that “went down” at a party – maybe a good incident, and maybe not.

This context could elicit any number of emotions and responses, like embarrassment, laughter, anger, or pride, and it is very possible that males and females tend to respond with unusual  emojis (at least compared to the “norms” established by our pilot survey) when faced with this situation. We did not account for these situational differences in our study, so further research is necessary to determine where these effects arise and to what degree.

There are a handful of other potential confounds in our study. For one, our survey did not include any control texts where no emoji was used at all. Such texts may have shed some light on the context confound previously discussed. Additionally, the survey only used 10 emojis, and the most popular ones at that. We have shown that these emojis have certain gender associations, but they are still the most commonly used emojis on Twitter, and certainly they are not used exclusively by males or exclusively by females. If anything, these emojis are used at comparable rates between genders. It might be more interesting to investigate emojis that are used mostly by one gender. One could find this information by scraping twitter and sorting emoji usage by gender.

Electronic paralinguistic cues are obviously not limited to emojis. Texters can make use of punctuation (periods, exclamation marks, etc.), caps, and makeshift emoticons, among other things, to compensate for the lack of visual and auditory information provided by ordinary paralinguistic cues. Further research might investigate how these other cues affect the gender connotation and tone perception of a text. Such research might also choose to look beyond gender and investigate how certain emojis are connected to perceived personality traits.

Ultimately, real-time written communication pervades our lives. For every sentence we speak, we text another. Somewhere in this mess of LOLs and BRBs and  ;)’s we must compensate for the unavailability of useful visual and auditory information, information we probably take for granted in everyday speech. Emojis, to this end, have become profoundly useful tools for providing additional communicative information, and we hope that our brief study has led you to agree. If not, all we can say is 😂.

References

Briton, N. J., & Hall, J. A. (1995). Beliefs about female and male nonverbal communication. Sex Roles, 32 (1-2), 79-90.

Appendix

In our experiment, we used three different text conversations, each having two versions: one that ends with a masculine emoji, and one that ends with a feminine emoji. We now show both versions of each text conversation.

Conversation 1: Lunch

 Conversation 2: Party

Conversation 3: Studying

Advertisements